New methods for estimating species trees from genome-scale data Tandy Warnow The University of Illinois Orangutan GorillaChimpanzee Human From the Tree of the Life Website, University of Arizona Phylogeny (evolutionary tree) Orangutan GorillaChimpanzee Human From the Tree of the Life Website, University of Arizona Sampling multiple genes from multiple species Incomplete Lineage Sorting (ILS) is a dominant cause of gene tree heterogeneity 1KP: Thousand Transcriptome Project 103 plant transcriptomes, single copy genes Next phase will be much bigger Wickett, Mirarab et al., PNAS 2014 G. Ka-Shu Wong U Alberta N. Wickett Northwestern J. Leebens-Mack U Georgia N. Matasci iPlant T. Warnow, S. Mirarab, N. Nguyen UT-Austin UT-Austin UT-Austin Challenge: Massive gene tree heterogeneity consistent with ILS Avian Phylogenomics Project E Jarvis, HHMI G Zhang, BGI Approx. 50 species, whole genomes, 14,000 loci Jarvis, Mirarab, et al., Science 2014 MTP Gilbert, Copenhagen S. Mirarab Md. S. Bayzid, UT-Austin UT-Austin T. Warnow UT-Austin Plus many many other people Major challenge: Massive gene tree heterogeneity consistent with incomplete lineage sorting This talk Gene tree heterogeneity due to incomplete lineage sorting, modelled by the multi-species coalescent (MSC) Statistically consistent estimation of species trees under the MSC, and the impact of gene tree estimation error ASTRAL (Bioinformatics 2014, 2015): coalescent-based species tree estimation method that has high accuracy on large datasets (1000 species and genes) Statistical binning (Science 2014) improving gene tree estimation, and hence species tree estimation Open questions This talk Gene tree heterogeneity due to incomplete lineage sorting, modelled by the multi-species coalescent (MSC) Statistically consistent estimation of species trees under the MSC, and the impact of gene tree estimation error ASTRAL (Bioinformatics 2014, 2015): coalescent-based species tree estimation method that has high accuracy on large datasets (1000 species and genes) Statistical binning (Science 2014) improving gene tree estimation, and hence species tree estimation Open questions Controversial! Gene trees inside the species tree (Coalescent Process) Present Past Courtesy James Degnan Gorilla and Orangutan are not siblings in the species tree, but they are in the gene tree. Incomplete Lineage Sorting (ILS) Confounds phylogenetic analysis for many groups: Hominids, Birds, Yeast, Animals, Toads, Fish, Fungi, etc. There is substantial debate about how to analyze phylogenomic datasets in the presence of ILS, focused around statistical consistency guarantees (theory) and performance on data. Statistical Consistency error Data ... Analyze separately Summary Method Two competing approaches gene 1 gene 2... gene k... Concatenation Species ... What about summary methods? ... What about summary methods? Techniques: Most frequent gene tree? Consensus of gene trees? Other? Statistically consistent under ILS? Coalescent-based summary methods: MP-EST (Liu et al. 2010): maximum pseudo-likelihood estimation of rooted species tree based on rooted triplet tree distribution YES NJst (Liu and Yu, 2011) - YES And others, including some newer methods (BUCKy-pop, ASTRAL, ASTRID, etc.) - YES Co-estimation methods: *BEAST (Heled and Drummond 2009): Bayesian co- estimation of gene trees and species trees YES Single-site methods (SVDquartets, METAL, SNAPP, and others) 1KP: Thousand Transcriptome Project 103 plant transcriptomes, single copy genes Next phase will be much bigger Wickett, Mirarab et al., PNAS 2014 G. Ka-Shu Wong U Alberta N. Wickett Northwestern J. Leebens-Mack U Georgia N. Matasci iPlant T. Warnow, S. Mirarab, N. Nguyen UT-Austin UT-Austin UT-Austin Challenges: Massive gene tree heterogeneity consistent with ILS Could not use MP-EST due to missing data (many gene trees could not be rooted) and large number of species 1KP: Thousand Transcriptome Project 103 plant transcriptomes, single copy genes Next phase will be much bigger Wickett, Mirarab et al., PNAS 2014 G. Ka-Shu Wong U Alberta N. Wickett Northwestern J. Leebens-Mack U Georgia N. Matasci iPlant T. Warnow, S. Mirarab, N. Nguyen UT-Austin UT-Austin UT-Austin Solution: New coalescent-based method ASTRAL ASTRAL is statistically consistent, polynomial time, and uses unrooted gene trees. ASTRAL and ASTRAL-2 Estimates the species tree from gene trees by finding the species tree that has the maximum quartet support, using dynamic programming Theorem: ASTRAL is statistically consistent under the MSC, even when solved in constrained mode (drawing bipartitions from the input gene trees) The constrained version of ASTRAL runs in polynomial time Open source software at https://github.com/smirarabhttps://github.com/smirarab Published in Bioinformatics 2014 and 2015 Used in Wickett, Mirarab et al. (PNAS 2014) and Prum, Berv et al. (Nature 2015) Used SimPhy, Mallo and Posada, 2015 Avian Phylogenomics Project E Jarvis, HHMI G Zhang, BGI Approx. 50 species, whole genomes, 14,000 loci Jarvis, Mirarab, et al., Science 2014 MTP Gilbert, Copenhagen S. Mirarab Md. S. Bayzid, UT-Austin UT-Austin T. Warnow UT-Austin Plus many many other people Major challenge: Massive gene tree heterogeneity consistent with incomplete lineage sorting Very poor resolution in the 14,000 gene trees (average bootstrap support 25%) Standard coalescent-based species tree estimation methods contradicted concatenation analysis and prior studies Statistical Consistency for summary methods error Data Data are gene trees, presumed to be randomly sampled true gene trees. Summary methods combine estimated gene trees, not true gene trees. Multiple studies show that summary methods can be less accurate than concatenation in the presence of high gene tree estimation error. Genome-scale data includes a range of markers, not all of which have substantial signal. Furthermore, removing sites due to model violations reduces signal. Some researchers also argue that gene trees should be based on very short alignments, to avoid intra-locus recombination. TYPICAL PHYLOGENOMICS PROBLEM: many poor gene trees Summary methods combine estimated gene trees, not true gene trees. Multiple studies show that summary methods can be less accurate than concatenation in the presence of high gene tree estimation error. Genome-scale data includes a range of markers, not all of which have substantial signal. Furthermore, removing sites due to model violations reduces signal. Some researchers also argue that gene trees should be based on very short alignments, to avoid intra-locus recombination. Gene tree estimation error: key issue in the debate Avian Phylogenomics Project E Jarvis, HHMI G Zhang, BGI Approx. 50 species, whole genomes, 14,000 loci Published Science 2014 MTP Gilbert, Copenhagen S. Mirarab Md. S. Bayzid, UT-Austin UT-Austin T. Warnow UT-Austin Plus many many other people Most gene trees had very low bootstrap support, suggestive of gene tree estimation error Avian Phylogenomics Project E Jarvis, HHMI G Zhang, BGI Approx. 50 species, whole genomes, 14,000 loci MTP Gilbert, Copenhagen S. Mirarab Md. S. Bayzid, UT-Austin UT-Austin T. Warnow UT-Austin Plus many many other people Solution: Statistical Binning Improves coalescent-based species tree estimation by improving gene trees (Mirarab, Bayzid, Boussau, and Warnow, Science 2014) Avian species tree estimated using Statistical Binning with MP-EST (Jarvis, Mirarab, et al., Science 2014) Gene Tree Estimation Error can be due to insufficient data error Data Data are sites in an alignment Note: Supergene trees computed using fully partitioned maximum likelihood Vertex-coloring graph with balanced color classes NP-hard problem; we used heuristic. Statistical binning vs. unbinned Datasets: 11-taxon strongILS datasets with 50 genes from Chung and An, Systematic Biology Binning produces bins with approximate 5 to 7 genes each Theorem 3 (PLOS One, Bayzid et al. 2015): Unweighted statistical binning pipelines are not statistically consistent under GTR+MSC As the number of sites per locus increase: All estimated gene trees converge to the true gene tree and have bootstrap support that converges to 1 (Steel 2014) For each bin, with probability converging to 1, the genes in the bin have the same tree topology (but can have different numeric parameters), and there is only one bin for any given tree topology For each bin, a fully partitioned maximum likelihood (ML) analysis of its supergene alignment converges to a tree with the common gene tree topology. As the number of loci increase: every gene tree topology appears with probability converging to 1. Hence as both the number of loci and number of sites per locus increase, with probability converging to 1, every gene tree topology appears exactly once in the set of supergene trees. It is impossible to infer the species tree from the flat distribution of gene trees! Fig 1. Pipeline for unbinned analyses, unweighted statistical binning, and weighted statistical binning. Bayzid MS, Mirarab S, Boussau B, Warnow T (2015) Weighted Statistical Binning: Enabling Statistically Consistent Genome-Scale Phylogenetic Analyses. PLoS ONE 10(6): e doi: /journal.pone Theorem 2 (PLOS One, Bayzid et al. 2015): WSB pipelines are statistically consistent under GTR+MSC Easy proof: As the number of sites per locus increase All estimated gene trees converge to the true gene tree and have bootstrap support that converges to 1 (Steel 2014) For every bin, with probability converging to 1, the genes in the bin have the same tree topology Fully partitioned GTR ML analysis of each bin converges to a tree with the common topology of the genes in the bin Hence as the number of sites per locus and number of loci both increase, WSB followed by a statistically consistent summary method will converge in probability to the true species tree. Q.E.D. Table 1. Model trees used in the Weighted Statistical Binning study. We show number of taxa, species tree branch length (relative to base model), and average topological discordance between true gene trees and true species tree. In the study we varied the number of genes and number of sites per gene. Dataset Species tree branch length scaling Average Discordance (%) Avian (48)2X35 Avian (48)1X47 Avian (48)0.5X59 Mammalian (37)2X18 Mammalian (37)1X32 Mammalian (37)0.5X54 10-taxonLower ILS"40 10-taxonHigher ILS"84 15-taxonHigh ILS"82 doi: /journal.pone t001 Species tree estimation error for MP-EST and ASTRAL, and also concatenation using ML, on avian simulated datasets: 48 taxa, moderately high ILS (AD=47%), 1000 genes, and varying gene sequence length. Impact of binning on species tree topology Bayzid et al., (2015). PLoS ONE 10(6): e Cumulative distribution of the bootstrap support values of true positive (left) and false positive (right) edges. If a curve for method X is above the curve for method Y, then X has higher BS for true positives and lower BS for false positives. Values in the shaded area indicate false positive branches with support at 75% or higher. Binning (weighted or unweighted) reduces the incidence of highly supported false positives, especially for the lower two ILS levels (1X and 2X branch lengths). Results are shown for 1000 genes with 500bp, on the avian simulated datasets. Impact of binning on bootstrap support Bayzid et al., (2015). PLoS ONE 10(6): e Weighted Statistical Binning: empirical WSB generally benign to highly beneficial for moderate to large datasets: Improves gene tree estimation Improves species tree topology Improves species tree branch length Reduces incidence of highly supported false positive branches Weighted Statistical Binning: empirical However, WSB can reduce accuracy under some conditions. Current simulations have only established this for model conditions that simultaneously have: Very small numbers of species (at most 10) Very high ILS (AD > 80%) Low bootstrap support for gene trees Most likely there are other conditions as well. Species tree estimation error for MP-EST and ASTRAL on 10-taxon datasets Bayzid MS, Mirarab S, Boussau B, Warnow T (2015). PLoS ONE 10(6): e Simphy Model Tree 200 genes with 100bp (GTRGAMMA) 10 replicates per condition Notes: Moderate ILS: binning neutral or beneficial using BS=50% Very high ILS: binning neutral for BS=50%, but increases MP-EST error with BS=75% AD=40% AD=84% Liu and Edwards, Comment in Science, October 2015 Attempted proof that WSB pipelines are statistically inconsistent for bounded number of sites per locus: The proof fails for multiple reasons, including the use of unpartitioned ML instead of fully partitioned ML Simulation study 5-taxon, strict molecular clock, very high ILS (AD=82%) performed WSB using unpartitioned ML instead of fully partitioned ML. erroneous (extopic) data in supergene alignments, biasing against WSB Our re-analysis of their data produced better results than they reported, but WSB did reduce accuracy on their data. Sketch of L&E argument For any sequence length L, there is a model species tree such that nearly all sites on nearly all genes evolve without any changes, and so nearly all gene trees have maximum bootstrap support below the threshold value. As the number of loci increase, the bins produced by WSB will have the same gene tree distribution as for the true species tree (or the deviation will not impact any downstream argument). On each bin, ML concatenation will converge to some tree that is not the species tree. (Hence, applying a coalescent-based method to these supergene trees will not produce the species tree, even as the number of loci increases.) Sketch of L&E argument For any sequence length L, there is a model species tree such that nearly all sites on nearly all genes evolve without any changes, and so nearly all gene trees have maximum bootstrap support below the threshold value. As the number of loci increase, the bins produced by WSB will have the same gene tree distribution as for the true species tree (or the deviation will not impact any downstream argument). On each bin, ML concatenation will converge to some tree that is not the species tree. (Hence, applying a coalescent-based method to these supergene trees will not produce the species tree, even as the number of loci increases.) Sketch of L&E argument For any sequence length L, there is a model species tree such that nearly all sites on nearly all genes evolve without any changes, and so nearly all gene trees have maximum bootstrap support below the threshold value. As the number of loci increase, the bins produced by WSB will have the same gene tree distribution as for the true species tree (or the deviation will not impact any downstream argument). On each bin, ML concatenation will converge to some tree that is not the species tree. (Hence, applying a coalescent-based method to these supergene trees will not produce the species tree, even as the number of loci increases.) Sketch of L&E argument For any sequence length L, there is a model species tree such that nearly all sites on nearly all genes evolve without any changes, and so nearly all gene trees have maximum bootstrap support below the threshold value. As the number of loci increase, the bins produced by WSB will have the same gene tree distribution as for the true species tree (or the deviation will not impact any downstream argument). On each bin, ML concatenation will converge to some tree that is not the species tree. (Hence, applying a coalescent-based method to these supergene trees will not converge to the species tree as the number of loci increases.) Liu and Edwards, Comment in Science, October 2015 Attempted proof that WSB pipelines are statistically inconsistent for bounded number of sites per locus: The proof fails for multiple reasons, including the use of unpartitioned ML instead of fully partitioned ML Simulation study 5-taxon, strict molecular clock, very high ILS (AD=82%) performed WSB using unpartitioned ML instead of fully partitioned ML. erroneous (extopic) data in supergene alignments, biasing against WSB Our re-analysis of their data produced better results than they reported, but WSB did reduce accuracy on their data. Liu and Edwards, Comment in Science, October 2015 Attempted proof that WSB pipelines are statistically inconsistent for bounded number of sites per locus: The proof fails for multiple reasons, including the use of unpartitioned ML instead of fully partitioned ML Simulation study 5-taxon, strict molecular clock, very high ILS (AD=82%) Our re-analysis of their data produced better results for statistical binning (both weighted and unweighted) than they reported, They performed WSB using unpartitioned ML instead of fully partitioned ML (biasing against statistical binning) They had erroneous (ectopic) data in their supergene alignments, biasing against statistical binning Figure of model tree from L&E, Science 9 October 2015: 171 This model tree fits into the category of conditions described in Bayzid et al. PLOS One 2015, in which WSB reduced accuracy (very small numbers of taxa, very high ILS). Liu and Edwards, Comment in Science, October 2015 Attempted proof that WSB pipelines are statistically inconsistent for bounded number of sites per locus: The proof fails for multiple reasons, including the use of unpartitioned ML instead of fully partitioned ML Simulation study 5-taxon, strict molecular clock, very high ILS (AD=82%) Our re-analysis of their data produced better results for statistical binning (both weighted and unweighted) than they reported They performed WSB using unpartitioned ML instead of fully partitioned ML (biasing against statistical binning) They had erroneous (ectopic) data in their supergene alignments, biasing against statistical binning Figure of model tree from L&E, Science 9 October 2015: 171 This model tree fits into the category of conditions described in Bayzid et al. PLOS One 2015, in which WSB reduced accuracy (very small numbers of taxa, very high ILS). Liu and Edwards, Comment in Science, October 2015 Attempted proof that WSB pipelines are statistically inconsistent for bounded number of sites per locus: The proof fails for multiple reasons, including the use of unpartitioned ML instead of fully partitioned ML Simulation study 5-taxon, strict molecular clock, very high ILS (AD=82%) Our re-analysis of their data produced better results for statistical binning (both weighted and unweighted) than they reported. They performed WSB using unpartitioned ML instead of fully partitioned ML (biasing against statistical binning). They had erroneous (ectopic) data in their supergene alignments, biasing against statistical binning Figure of model tree from L&E, Science 9 October 2015: 171 This model tree fits into the category of conditions described in Bayzid et al. PLOS One 2015, in which WSB reduced accuracy (very small numbers of taxa, very high ILS). Liu and Edwards, Comment in Science, October 2015 Attempted proof that WSB pipelines are statistically inconsistent for bounded number of sites per locus: The proof fails for multiple reasons, including the use of unpartitioned ML instead of fully partitioned ML Simulation study 5-taxon, strict molecular clock, very high ILS (AD=82%) Our re-analysis of their data produced better results for statistical binning (both weighted and unweighted) than they reported. They performed WSB using unpartitioned ML instead of fully partitioned ML (biasing against statistical binning) They had erroneous (ectopic) data in their supergene alignments, biasing against statistical binning. Figure of model tree from L&E, Science 9 October 2015: 171 This model tree fits into the category of conditions described in Bayzid et al. PLOS One 2015, in which WSB reduced accuracy (very small numbers of taxa, very high ILS). Liu and Edwards, Comment in Science, October 2015 Attempted proof that WSB pipelines are statistically inconsistent for bounded number of sites per locus: The proof fails for multiple reasons, including the use of unpartitioned ML instead of fully partitioned ML Simulation study 5-taxon, strict molecular clock, very high ILS (AD=82%) Figure of model tree from L&E, Science 9 October 2015: 171 Liu and Edwards, Comment in Science, October 2015 Attempted proof that WSB pipelines are statistically inconsistent for bounded number of sites per locus: The proof fails for multiple reasons, including the use of unpartitioned ML instead of fully partitioned ML Simulation study 5-taxon, strict molecular clock, very high ILS (AD=82%) Figure of model tree from L&E, Science 9 October 2015: 171 Mirarab et al. response (Science, October 2015): Our re-analysis of their data (with correct supergene alignments) shows that WSB reduces accuracy but not by as much as they report. Our analyses of slightly larger datasets with the same properties (pectinate, very high ILS, strict clock) showed WSB neutral to beneficial. Liu and Edwards, Comment in Science, October 2015 Attempted proof that WSB pipelines are statistically inconsistent for bounded number of sites per locus: The proof fails for multiple reasons, including the use of unpartitioned ML instead of fully partitioned ML Simulation study 5-taxon, strict molecular clock, very high ILS (AD=82%) Figure of model tree from L&E, Science 9 October 2015: 171 All the conditions in which WSB has been shown to reduce accuracy have the following properties: High ILS (AD > 80%) Small numbers of taxa (at most 10) Low bootstrap support on gene trees and most also obeyed the strict molecular clock. Question: Do any summary methods converge to the species tree as the number of loci increase, but where each locus has only a constant number of sites? Answers: Roch & Warnow, Syst Biol, March 2015: Strict molecular clock: Yes for some new methods, even for a single site per locus No clock: Unknown for all methods, including MP-EST, ASTRAL, etc. L&E ask a good question: performance on bounded number of sites! Question: Do any summary methods converge to the species tree as the number of loci increase, but where each locus has only a constant number of sites? Answers: Roch & Warnow, Syst Biol, March 2015: Strict molecular clock: Yes for some new methods, even for a single site per locus No clock: Unknown for all methods, including MP-EST, ASTRAL, etc. S. Roch and T. Warnow. "On the robustness to gene tree estimation error (or lack thereof) of coalescent-based species tree methods", Systematic Biology, 64(4): , 2015, (PDF)(PDF) L&E ask a good question: performance on bounded number of sites! For any positive integer L, if all loci have at most L sites, then coalescent-based summary methods (e.g., MP-EST, ASTRAL, NJst, ASTRID, etc.) cannot be statistically consistent under the MSC. Comments: All current proofs of statistical consistency for standard summary methods assume perfect gene trees, and so inherently assume that sequence lengths for all genes go to infinity. The challenge is that for some model gene trees, ML can be biased towards the wrong tree topology for small enough L (think long branch attraction, Felsenstein Zone). Conjecture For any positive integer L, if all loci have at most L sites, then coalescent-based summary methods (e.g., MP-EST, ASTRAL, NJst, ASTRID, etc.) cannot be statistically consistent under the MSC. Comments: All current proofs of statistical consistency for standard summary methods assume perfect gene trees, and so inherently assume that sequence lengths for all genes go to infinity. The challenge is that for some model gene trees, ML can be biased towards the wrong tree topology for small enough L (think long branch attraction, Felsenstein Zone). Conjecture For any positive integer L, if all loci have at most L sites, then coalescent-based summary methods (e.g., MP-EST, ASTRAL, NJst, ASTRID, etc.) cannot be statistically consistent under the MSC. Comments: All current proofs of statistical consistency for standard summary methods assume perfect gene trees, and so inherently assume that sequence lengths for all genes go to infinity. The challenge is that for some model gene trees, ML can be biased towards the wrong tree topology for small enough L (think long branch attraction, Felsenstein Zone). Conjecture For any positive integer L, if all loci have at most L sites, then WSB pipelines cannot be statistically consistent under the MSC. Comments: Might well be true! The proof will not be easy. Rephrasing L&E Technical Comment as a conjecture For any positive integer L, if all loci have at most L sites, then WSB pipelines cannot be statistically consistent under the MSC. Comments: Open Will be hard to settle either way Rephrasing L&E Technical Comment as a conjecture Consistency first kind: both number of loci and number of sites go to infinity Consistency second kind: number of loci goes to infinity, number of sites bounded by L (arbitrary constant) Table from PLOS Currents, Warnow 2015 * * No theoretical difference between MP-EST, ASTRAL, and WSB (according to current knowledge) * Future Directions Better coalescent-based summary methods (that are more robust to gene tree estimation error) Better techniques for estimating gene trees given multi-locus data, or for co-estimating gene trees and species trees Better theory about robustness to gene tree estimation error (or lack thereof) for coalescent- based summary methods Better single site methods Future Directions Better coalescent-based summary methods (that are more robust to gene tree estimation error) Better techniques for estimating gene trees given multi-locus data, or for co-estimating gene trees and species trees Better theory about robustness to gene tree estimation error (or lack thereof) for coalescent- based summary methods Better single site methods Future Directions Better coalescent-based summary methods (that are more robust to gene tree estimation error) Better techniques for estimating gene trees given multi-locus data, or for co-estimating gene trees and species trees Better theory about robustness to gene tree estimation error (or lack thereof) for coalescent- based summary methods Better single site methods Future Directions Better coalescent-based summary methods (that are more robust to gene tree estimation error) Better techniques for estimating gene trees given multi-locus data, or for co-estimating gene trees and species trees Better theory about robustness to gene tree estimation error (or lack thereof) for coalescent- based summary methods Better single site methods Acknowledgments Mirarab et al., Science 2014 (Statistical Binning) Roch and Warnow, Systematic Biology 2014 (Points of View) Bayzid et al., Science 2015 (Response to Liu and Edwards Comment) Mirarab and Warnow, Bioinformatics 2015 (ASTRAL-2) Warnow PLOS Currents: Tree of Life 2014 (concatenation analysis) Papers available atASTRAL and statistical binning software at https://github.com/smirarabhttps://github.com/smirarab Funding: NSF, David Bruton Jr. Centennial Professorship, TACC (Texas Advanced Computing Center), Grainger Foundation, and HHMI (to SM). Comparing Binned and Un-binned MP-EST on the Avian Dataset Binned MP-EST is largely consistent with the ML concatenation analysis. The trees presented in Science 2014 were the ML concatenation and Binned MP-EST Bayzid et al., (2015). PLoS ONE 10(6): e Species tree estimation error for MP-EST and ASTRAL on 15-taxon datasets Bayzid MS, Mirarab S, Boussau B, Warnow T (2015). PLoS ONE 10(6): e Model Tree: Very high ILS: AD = 82% Strict molecular clock GTR+Gamma sequence evolution (Indelible) 10 replicates per condition Notes: BS-75% often improved accuracy (p=0.04) BS=50% sometimes reduced accuracy, but differences were not statistically significant. MP-EST more accurate than ASTRAL on these data. Research Questions Why does concatenation using ML produce such good accuracy under many conditions? Why does statistical binning improve accuracy under many conditions? What kind of method should be used to compute a species tree, or does this depend on the estimated amount of ILS and gene tree accuracy in the dataset? What is a biologically reasonable amount of ILS? How can we use multiple loci to help improve the estimation of individual gene trees? (Note: co-estimation under the MSC is very powerful, but current methods are not able to analyze even moderate-sized datasets.) Bayzid et al. (PLOS One 2015): Thus, statistical binning seems to be beneficial when both ILS level and gene tree bootstrap support are not too high, will be neutral when bootstrap support values are high (so little or no binning occurs), but can be detrimental when ILS levels are extremely high but gene tree bootstrap support is low enough that binning occurs. Thus, one consequence of this study is the suggestion that when ILS levels are very high and the average gene tree bootstrap support is low, then either statistical binning should not be used, or it should be used in a very conservative fashionwith the parameter B set very low. Mirarab et al. (Science 2015): Studies on 10- and 15-taxon datasets similar to the Liu and Edwards 5-taxon datasets showed binning was neutral to beneficial. Hence dataset size also seems to be relevant (i.e., binning might be potentially detrimental on very small datasets). When not to use statistical binning Summary Unpartitioned concatenation using maximum likelihood is statistically inconsistent under the MSC (Roch and Steel 2014, see Warnow PLOS Currents 2015). However, concatenation can be highly accurate (and even more accurate than the best coalescent-based methods currently available) under low enough ILS. Concatenation is controversial. Many coalescent-based summary methods (e.g., MP-EST and ASTRAL) converge In probability to the true species tree as the number of gene trees increase. However, all proofs to date have assumed error-free gene trees, and gene tree estimation error clearly impacts species tree estimation accuracy (and not just bootstrap support). Some new summary methods can have excellent accuracy even on large datasets (e.g., ASTRAL-2). However, summary methods are controversial. Statistical binning (Mirarab et al. Science 2014) and weighted statistical binning (Bayzid et al., PLOS One) often (but not always) improve gene tree estimation, and hence coalescent-based species tree estimation from multiple genes. However, statistical binning is controversial. Summary Unpartitioned concatenation using maximum likelihood is statistically inconsistent under the MSC (Roch and Steel 2014, see Warnow PLOS Currents 2015). However, concatenation can be highly accurate (and even more accurate than the best coalescent-based methods currently available) under low enough ILS. Concatenation is controversial. Many coalescent-based summary methods (e.g., MP-EST and ASTRAL) converge In probability to the true species tree as the number of gene trees increase. However, all proofs to date have assumed error-free gene trees, and gene tree estimation error clearly impacts species tree estimation accuracy (and not just bootstrap support). Some new summary methods can have excellent accuracy even on large datasets (e.g., ASTRAL-2). However, summary methods are controversial. Statistical binning (Mirarab et al. Science 2014) and weighted statistical binning (Bayzid et al., PLOS One) often (but not always) improve gene tree estimation, and hence coalescent-based species tree estimation from multiple genes. However, statistical binning is controversial. Summary Unpartitioned concatenation using maximum likelihood is statistically inconsistent under the MSC (Roch and Steel 2014, see Warnow PLOS Currents 2015). However, concatenation can be highly accurate (and even more accurate than the best coalescent-based methods currently available) under low enough ILS. Concatenation is controversial. Many coalescent-based summary methods (e.g., MP-EST and ASTRAL) converge In probability to the true species tree as the number of gene trees increase. However, all proofs to date have assumed error-free gene trees, and gene tree estimation error clearly impacts species tree estimation accuracy (and not just bootstrap support). Some new summary methods can have excellent accuracy even on large datasets (e.g., ASTRAL-2). However, summary methods are controversial. Statistical binning (Mirarab et al. Science 2014) and weighted statistical binning (Bayzid et al., PLOS One) often (but not always) improve gene tree estimation, and hence coalescent-based species tree estimation from multiple genes. However, statistical binning is controversial.